The Prompt Enhancer analyses a reference image and rewrites your prompt with concrete cues — camera angle, lens, lighting, environment, mood — so the generator has more to work with.
When to use it
Use Prompt Enhancer when:
- The reference is rich (a film still, a moodboard pin) and you want the generator to match its style.
- Your initial prompt is short (“woman walking through forest”) and you want the model to fill in concrete detail.
- You are iterating on a generation and want the prompt to evolve along with the references.
How to run it
- In the image generation panel, attach at least one reference image to the Input Image slot.
- Click Prompt Enhancer.
- The enhancer reads the reference, your current prompt, and the selected model, and proposes a rewritten prompt with concrete additions (camera, lighting, environment).
- The proposed prompt replaces the text in the prompt box. Edit further if you want.
What gets added
A typical enhancement adds:
- Camera details — focal length, depth of field, shot type (close-up, wide, low angle).
- Lighting — direction, colour temperature, mood (cinematic, harsh, ambient).
- Environment — time of day, weather, surroundings consistent with the reference.
- Style references — film stock, photographer name, art movement when the reference suggests one.
The enhancer does not add fictional brand names, copyrighted characters, or person names you did not provide.
Which model the enhancer uses
The enhancer runs on the LLM you have configured — by default the same model you pick in the agent dock (GPT, Claude, or Gemini). It needs a key for at least one of those providers; see Configuring API Keys.
Cost
The enhancement call is a single LLM round-trip with a small context. Cost is typically a fraction of a cent and is included in the cost estimate shown next to the Generate button.
Leave a Reply